1. Field of the Invention
The present invention relates generally to a photosensor signal processing apparatus for suppressing a fixed pattern noise resulting from variations (unevenness) in characteristics of pixels of a metal-oxide-semiconductor (MOS) image sensor. More particularly, it relates to a photosensor signal processing apparatus for correcting the output of a MOS image sensor having an inflection point at which the output of each pixel changes from a linear function to a logarithmic function of incident illumination, and vice versa.
2. Description of the Related Art
In order to suppress a fixed pattern noise resulting from variations (unevenness) in characteristics of pixels of a MOS image sensor, one conventional approach has been directed toward improvements in the device fabrication process such that all of the pixel-forming devices of the MOS image sensor have uniform output characteristics.
In another conventional approach, a measurement is made in advance to determine characteristics of a pattern noise of the MOS image sensor. The fixed pattern noise is suppressed through a correction made subsequently on the basis of the predetermined pattern noise characteristics.
Japanese Patent Publication No. SHO 61-14702 discloses a method of correcting output variations between pixels of an image sensor. The output of this image sensor varies linearly with incident illumination. To correct the pixel-to-pixel output variation of the image sensor, individual output values "Did" from the pixels in a dark state and individual output values "Diw" from the same pixels in a bright state under uniform illumination are previously stored in a memory. An output value "D1" of each pixel, which is produced in response to light and shade of an image picked up by the image sensor, is corrected by arithmetic operation achieved in accordance with the formula: A(Di-Did)/Diw-Did) where A is a constant. With this arithmetic operation, the pixel-to-pixel output variation in the dark state and the pixel-to-pixel sensitivity variation (variations in the gradients of linear output characteristic curves) are corrected.
A solid-state imaging system disclosed in Japanese Patent Laid-open Publication No. HEI 5-30350 includes a charge-coupled device (CCD) having a mosaic of tiny pixels whose outputs vary logarithmically with incident illumination. To correct the pixel-to-pixel output variation, individual imaging data of the pixels are previously collected under uniform illumination, then stored in a memory. These pre-stored imaging data are subtracted pixel by pixel from individual imaging data of the pixels collected during actual imaging operation. With this subtracting operation, the pixel-to-pixel logarithmic output characteristic variation is corrected.
The output variation correcting method disclosed in the first-mentioned Japanese Publication cannot deal with the correction of variations in output of pixels having a logarithmic property. The solid-state imaging system shown in the second-mentioned Japanese Publication can achieve correction of the pixel-to-pixel logarithmic output characteristic variation. However, due to the subtracting operation, the intensity of corrected sensor output of the imaging system is lower than that of sensor output corresponding to actual incident illumination. The sensitivity of the overall imaging system goes down, accordingly.
Japanese Patent Laid-open Publication No. HEI 10-90058 discloses a MOS image sensor designed to produce a sensor output having both a linear function region and a logarithmic function region with respect to incident illumination. A problem associated with this known MOS image sensor is that at an inflecting point where the sensor output changes from the linear function to the logarithmic function, and vice versa, an output error resulting from the pixel-to-pixel characteristic variation (variations in characteristic of MOS transistors and photodiodes forming the pixels) is likely to increase.